Papers by Parag Agrawal
SCULPT: Systematic Tuning of Long Prompts (2025.acl-long)
Copied to clipboard
Shanu Kumar, Akhila Yesantarao Venkata, Shubhanshu Khandelwal, Bishal Santra, Parag Agrawal, Manish Gupta
| Challenge: | Existing methods for prompt optimization struggle with longer, more complex ones, often risking information loss and being sensitive to small perturbations. |
| Approach: | They propose a framework that treats prompt optimization as a hierarchical tree refinement problem and uses a Critic-Actor framework to generate reflections and apply actions to refine the prompt. |
| Outcome: | The proposed framework produces more stable and interpretable prompt modifications, ensuring better generalization across tasks. |
SAGE: A Generic Framework for LLM Safety Evaluation (2025.emnlp-industry)
Copied to clipboard
| Challenge: | Current safety evaluation methodologies focus on single-turn interactions with generic policies, failing to capture conversational dynamics of real-world usage and application-specific harms. |
| Approach: | They propose a framework for customized and dynamic harm evaluations that employs prompted adversarial agents with diverse personalities based on the Big Five model. |
| Outcome: | The proposed framework enables system-aware multi-turn conversations that adapt to target applications and harm policies. |
Unified Semantic Parsing with Weak Supervision (P19-1)
Copied to clipboard
Priyanka Agrawal, Ayushi Dalmia, Parag Jain, Abhishek Bansal, Ashish Mittal, Karthik Sankaranarayanan
| Challenge: | Semantic parsing over multiple knowledge bases requires high-quality annotations of (utterance, program) pairs. |
| Approach: | They propose a framework to build a unified multi-domain enabled semantic parser with weak supervision. |
| Outcome: | The proposed model improves performance by 20% on the Overnight dataset. |
Navigating the Cultural Kaleidoscope: A Hitchhiker’s Guide to Sensitivity in Large Language Models (2025.naacl-long)
Copied to clipboard
Somnath Banerjee, Sayan Layek, Hari Shrawgi, Rajarshi Mandal, Avik Halder, Shanu Kumar, Sagnik Basu, Parag Agrawal, Rima Hazra, Animesh Mukherjee
| Challenge: | Cultural harm arises when LLMs misrepresent or normalize values, identities, and practices in ways that conflict with the norms of diverse cultural groups. |
| Approach: | They propose a cultural harm test dataset and a preference dataset to assess model outputs across different cultural contexts. |
| Outcome: | The proposed model improves model behavior significantly reducing the likelihood of generating culturally insensitive or harmful content. |
Explanations for CommonsenseQA: New Dataset and Models (2021.acl-long)
Copied to clipboard
Shourya Aggarwal, Divyanshu Mandowara, Vishwajeet Agrawal, Dinesh Khandelwal, Parag Singla, Dinesh Garg
| Challenge: | a dataset called CommonsenseQA (CQA) was recently released to advance the research on common-sense question answering (QA) |
| Approach: | They propose to retrieve and generate explanations for a given question, correct answer choice, incorrect answer choices tuple from a dataset called CommonsenseQA. |
| Outcome: | The proposed model beats baseline model by 100% in F1 score and similarity score of 61.9 . |
LLM Safety for Children (2025.naacl-industry)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are increasingly impacting children through education, toys, and therapy, offering benefits like improved mental health and parental controls. |
| Approach: | They propose a comprehensive approach to evaluating LLM safety specifically for children by listing potential risks that children may encounter when using LLM-powered applications. |
| Outcome: | The proposed model bridges the gap in child safety literature across various fields. |
Enhancing Zero-shot Chain of Thought Prompting via Uncertainty-Guided Strategy Selection (2025.coling-main)
Copied to clipboard
Shanu Kumar, Saish Mendke, Karody Lubna Abdul Rahman, Santosh Kurasa, Parag Agrawal, Sandipan Dandapat
| Challenge: | Existing methods for chain-of-thought (CoT) prompting are limited by handcrafted demonstrations and trigger phrases are prone to inaccuracies. |
| Approach: | They propose a method that generates rationales using a trigger phrase to select effective demonstrations without accessing model parameters. |
| Outcome: | The proposed method outperforms existing methods across four reasoning benchmarks and is robust and scalable. |